netdev - Re: [PATCH net v4 1/2] net: sched: tbf: fix calculation of max

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 3 Dec 2013 17:47:45 +0800
From:	Yang Yingliang <yangyingliang@...wei.com>
To:	Eric Dumazet <eric.dumazet@...il.com>
CC:	<davem@...emloft.net>, <netdev@...r.kernel.org>,
	<brouer@...hat.com>, <jpirko@...hat.com>, <jbrouer@...hat.com>
Subject: Re: [PATCH net v4 1/2] net: sched: tbf: fix calculation of max_size

On 2013/12/3 16:12, Eric Dumazet wrote:
> On Tue, 2013-12-03 at 15:44 +0800, Yang Yingliang wrote:
>> On 2013/12/3 12:59, Eric Dumazet wrote:
>>> On Tue, 2013-12-03 at 11:26 +0800, Yang Yingliang wrote:
>>>
>>>> +	for (max_size = 0; max_size < MAX_PKT_LEN; max_size++)
>>>> +		if (psched_l2t_ns(&q->rate, max_size) > q->buffer)
>>>> +			break;
>>>> +	if (--max_size <= 0)
>>>> +		goto unlock_done;
>>>> +
>>>
>>> This seems dubious. With your new code, max_size < 65536
>>>
>>> Prior code had :
>>>
>>> for (n = 0; n < 256; n++)
>>>     if (rtab->data[n] > qopt->buffer)
>>>         break;
>>> max_size = (n << qopt->rate.cell_log) - 1;
>>>
>>> So we could have much bigger max_size.
>>>
>>> The reason I ask is that its possible to have qdisc_pkt_len(skb) being
>>> bigger than 65536, for TCP packets with low MSS value.
>>>
>>
>> Hmmm, if qdisc_pkt_len(skb) is bigger than 65536, skb_is_gso(skb) is true,
>> it will go into tbf_segment(). If I am wrong, please point me out, thanks!
>>
>> BTW, 65536 is suggested by Jesper, I'm a little uncertain about it. He is, too.
>> Do you or some other developers have stronger opinions on this?
> 
> We do not want to go to tbf_segment() if we programmed tbf to allow TSO
> packets of 68.000 bytes being sent without being segmented.

I mean a 64KB TSO packet's "qdisc_pkt_len(skb)" is bigger than 65536, but it will go
to tbf_segment(), so the TSO packet of 64KB can be enqueued.

> 
> TSO packet of 64KB -> about 45 frames if MSS=1448, 45*1514 = 68130 bytes
> 
> Have you tried to use TBF on a 10 Gbps link, say with one TCP flow ?

Yes, I had tried it with intel82599 nic.


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html